Moving human indication apparatus using doppler radar sensor and method thereof

Abstract

The present invention relates to a device and method for identifying a moving human using a Doppler radar sensor. According to the present invention, a moving human recognition device based on a Doppler radar comprises: a signal acquisition unit that acquires a Doppler spectrum for each target detected in each frame from a radar reception signal; a signal processing unit that extracts a plurality of feature values ​​related to the movement of a target by analyzing the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of each frame for the target, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude; and a human determination unit that identifies the presence or absence of a moving human using the plurality of feature values. According to the present invention, based on a simple and miniaturizable Doppler radar, it is possible to clearly distinguish whether a currently detected moving target is human or not, and based on this, accurately determine the presence of a human.

Description

Technology Field

[0001] The present invention relates to a device and method for identifying a moving human using a Doppler radar sensor, and more specifically, to a device and method for identifying a moving human using a Doppler radar sensor capable of accurately detecting a moving human. Background Technology

[0002] With the recent establishment of various smart environments such as smart cities, smart buildings, smart homes, and smart surveillance, the importance of human detection sensors is being highlighted.

[0003] Conventional camera sensors detect humans based on images, similar to the human eye, but they are sensitive to external environments and lighting, and consumers often object to them, particularly due to privacy concerns. Unlike other sensors, radar sensors have the advantage of being able to measure a person's biosignals non-contactually.

[0004] Currently, the Frequency Modulated Continuous Wave (FMCW) method, which enables the detection of distance, speed, and angle of multiple objects, is widely used in automotive radar sensors. Furthermore, research is also being conducted to distinguish various types of objects based on FMCW radar sensors.

[0005] However, depending on the various smart applications, there are cases where low-cost, small sensors are required rather than high-performance functions like automotive radar. Furthermore, to implement simple functions that merely distinguish whether an object is human or not, rather than handling multiple objects, a CW-based Doppler radar sensor may be more suitable than an FMCW-based one.

[0006] Doppler radar has a much simpler circuit and can be implemented with low power consumption compared to FMCW radar. However, it cannot measure distance, and the received signal only extracts the Doppler signal of a moving object. Nevertheless, if it is possible to determine whether a moving object is human or not based on the detected signal using Doppler radar, its range of applications would be very diverse.

[0007] As a related prior art study, Korean Published Patent No. 10-2023-0080737 by the present applicant discloses a technology for distinguishing whether a moving object is human or not by extracting features from a signal of a Doppler radar sensor, namely the number of scattered points in the current frame, the difference in the number of scattered points in consecutive frames, and the amount of change in signal magnitude in consecutive frames.

[0008] However, since the magnitude of the radar received signal is inversely proportional to the fourth power of the distance (signal attenuates significantly with distance), the third characteristic, the change in signal magnitude, must be normalized through the detected distance. In this regard, Figure 1 is a diagram showing the attenuation of the radar received signal magnitude according to distance. As shown in Figure 1, it can be seen that the magnitude of the received signal attenuates significantly with distance, and since the third characteristic represents the rate of change in signal magnitude between the current frame and the previous frame, in order to measure the accurate rate of change in signal, the signal attenuation characteristic according to the distance of a moving object must be reflected.

[0009] In the case of the general FMCW method, signal magnitude correction is possible using measured distance values; however, since the CW Doppler radar method cannot measure distance values, the aforementioned prior study calculated the third feature vector by utilizing the approximation formula mentioned in the mathematical equation.

[0010] However, when conducting actual performance tests, the error in the approximation value is severe when there is a moving object at 2 to 3 meters or less, making it difficult to use the third feature vector for object detection. In addition, as shown in Figure 1, it can be observed that the reception signal attenuation drops sharply up to 2 to 3 meters, but then drops gradually thereafter.

[0011] Furthermore, unlike automobiles, humans have the characteristic that the position of the reflected scattering point continuously changes frame by frame, and the detected velocity value can vary slightly depending on the position of that scattering point. However, prior studies did not consider this characteristic at all, which can lead to errors where moving humans are mistaken for automobiles.

[0012] Therefore, a technique is required to further complement the existing proposed Doppler radar sensor-based moving human detection algorithm.

[0013] The technology forming the background of the present invention is disclosed in Korean Published Patent No. 10-2023-0080737 (published June 7, 2023). The problem to be solved

[0014] The present invention aims to provide a moving human identification device and method using a Doppler radar sensor capable of distinguishing whether a moving object is a human from a received signal of a Doppler radar and accurately determining the presence of a human based thereon. means of solving the problem

[0015] The present invention relates to a moving human recognition device based on a Doppler radar, comprising: a signal acquisition unit that acquires a Doppler spectrum for each target detected in each frame from a radar reception signal; a signal processing unit that extracts a plurality of feature values ​​related to the movement of a target by analyzing the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of each frame for the target, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude; and a human determination unit that identifies the presence or absence of a moving human using the plurality of feature values.

[0016] Additionally, the signal processing unit can calculate a first feature value (M), which is the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of the current frame for the target; a second feature value (N), which is the difference between the first feature value (M') of the previous frame and the first feature value (M); a third feature value (K), which is the rate of change of the maximum signal magnitude of the current frame relative to the maximum signal magnitude of the previous frame; and a fourth feature value (L), which is the rate of change of the velocity value corresponding to the maximum signal magnitude of the current frame relative to the velocity value corresponding to the maximum signal magnitude of the previous frame.

[0017] In addition, the above third feature value (K) can be calculated by the following mathematical formula.

[0018]

[0019] Here, A is the maximum signal magnitude in the Doppler spectrum of the current frame, A' is the maximum signal magnitude in the Doppler spectrum of the previous frame, V is the velocity of the target detected in the current frame, and T up represents the time interval between the above frames.

[0020] In addition, the human determination unit can identify the presence or absence of a moving human by operating a first mode that utilizes all of the first to fourth feature values ​​according to preset conditions, or by operating a second mode that utilizes the first, second, and fourth feature values ​​excluding the third feature value.

[0021] Additionally, the human determination unit may operate by driving a first mode for a short-range detection area within a preset threshold distance to identify the presence or absence of a moving human using the first, second, and fourth feature values ​​excluding the third feature value, and by driving a second mode for a long-range detection area beyond the threshold distance to identify the presence or absence of a moving human using all of the first to fourth feature values.

[0022] In addition, the human determination unit can determine whether a moving human exists by applying the first to fourth feature values ​​or the first, second, and fourth feature values ​​to a pre-trained machine learning model.

[0023] And, the present invention includes a method for recognizing a moving human based on a Doppler radar, comprising: a step of acquiring a Doppler spectrum for each target detected in each frame from a radar reception signal; a step of extracting a plurality of feature values ​​related to the movement of a target by analyzing the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of each frame for the target, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude; and a step of identifying the presence or absence of a moving human using the plurality of feature values. Effects of the invention

[0024] According to the present invention, based on a simple and miniaturizable Doppler radar, it is possible to clearly distinguish whether a currently detected moving target is human or not, and based on this, accurately determine the presence of a human. Brief explanation of the drawing

[0025] Figure 1 is a diagram showing the attenuation of the magnitude of the radar received signal according to distance. FIG. 2 is a diagram showing the configuration of a mobile human identification device using a Doppler radar sensor according to an embodiment of the present invention. Figure 3 is a diagram that explains the configuration of the signal processing unit shown in Figure 2 in more detail. Figure 4 is a diagram showing an example of the Doppler spectra of the immediate past frame and the current frame for a moving pedestrian in an embodiment of the present invention. FIG. 5 is a diagram showing an example of the Doppler spectrum of the immediate past frame and the current frame for a moving vehicle in an embodiment of the present invention. FIG. 6 is a diagram illustrating a method for identifying moving humans using a Doppler radar sensor according to an embodiment of the present invention. Specific details for implementing the invention

[0026] Then, with reference to the attached drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily implement the invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification have been given similar reference numerals.

[0027] Throughout the specification, when a part is described as being "connected" to another part, this includes not only cases where they are "directly connected," but also cases where they are "electrically connected" with other components interposed between them. Furthermore, when a part is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0028] FIG. 2 is a diagram showing the configuration of a mobile human identification device using a Doppler radar sensor according to an embodiment of the present invention.

[0029] As shown in FIG. 2, a mobile human identification device (100) using a Doppler radar sensor according to an embodiment of the present invention includes a signal acquisition unit (110), a signal processing unit (120), and a human determination unit (130). Here, the operation of each unit (110 to 130) and the data flow between each unit can be controlled by a control unit (not shown).

[0030] The mobile human identification device (100) according to an embodiment of the present invention may be implemented by being included in a radar device (sensor) and may be operated in a form connected to the radar device.

[0031] The radar device used in the embodiment of the present invention corresponds to a CW-type Doppler radar (CW radar). The radar device can amplify an RF signal and transmit it through a transmitting antenna, and amplify a signal received from a receiving antenna and mix it in a mixer to extract and provide an IQ signal.

[0032] The mobile human identification device (100) can generate a Doppler spectrum through frequency analysis based on an IQ signal obtained from a radar device and extract four feature vectors. In the following embodiment of the present invention, the human being to be detected and recognized may include a mobile human (pedestrian) existing within the detection area, and other moving objects (e.g., vehicles) other than the mobile human may be recognized.

[0033] First, the signal acquisition unit (110) acquires the Doppler spectrum for each target detected in each frame from the radar reception signal. The signal acquisition unit (110) can generate the Doppler spectrum for each frame using the radar reception signal acquired through the Doppler radar sensor and transmit it to the signal processing unit (120).

[0034] The signal processing unit (120) analyzes the number of Doppler bins greater than a reference value, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude in the Doppler spectrum of each frame for the target, and extracts multiple feature values ​​(M, N, K, L) related to the target. Here, the feature values ​​may correspond to feature values ​​related to the movement (motion) of the target.

[0035] Figure 3 is a diagram that explains the configuration of the signal processing unit shown in Figure 2 in more detail.

[0036] Specifically, the signal processing unit (120) can calculate, as shown in FIG. 3, a first feature value (M) which is the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of the current frame for the target, a second feature value (N) which is the difference between the first feature value (M') of the previous frame and the first feature value (M), a third feature value (K) which is the rate of change of the maximum signal size of the current frame relative to the maximum signal size of the previous frame, and a fourth feature value (L) which is the rate of change of the speed value corresponding to the maximum signal size of the current frame relative to the speed value corresponding to the maximum signal size of the previous frame.

[0037] The first and second characteristic values ​​are explained in more detail as follows.

[0038] The first feature value (M) corresponds to an indicator (feature vector 1) regarding whether the Doppler spectrum of the current frame for the target is sufficiently widely distributed, and the second feature value (N) corresponds to an indicator (feature vector 2) regarding whether there is a large difference between the distribution of the Doppler spectrum of the current frame and the Doppler spectrum of the previous frame.

[0039] If the number of Doppler bins in the current frame's Doppler spectrum that have magnitudes greater than the reference value is large (the larger the value of M), it corresponds to the spectrum being spread out in a bell shape along the frequency axis. In this case, the probability that the object is a human (pedestrian) increases.

[0040] In addition, in the case of pedestrians, the spectrum appears to spread out in a bell shape with respect to the frequency axis, whereas for vehicles, it appears to have a sharp shape in a narrow band. Therefore, it is possible to distinguish whether it is a human or a vehicle by using the shape of the spectral distribution (first feature value; M) (degree of spectrum spread) within the current Doppler spectrum.

[0041] However, in some cases, even in the case of a human (pedestrian), the spectrum may appear sharp like a vehicle, so the difference in the Doppler pattern between the current frame and the previous frame's Doppler spectrum (second feature value; N) can be additionally analyzed and utilized to reduce errors in object determination. That is, regardless of the first feature value (M), the probability that the object is a human increases even when the second feature value (N) is large. Here, an example of the calculation of the values ​​of M and N is disclosed in detail in Registered Patent No. 10-2227393 by the applicant.

[0042] Additionally, the signal processing unit (120) can calculate a third feature value (K), which is the rate of change of the maximum signal size of the current frame relative to the maximum signal size of the previous frame.

[0043] The third feature value corresponds to an indicator (feature vector 3) representing the rate of change in signal magnitude over time for the target. Even when the third feature value (K) is high, the probability that the object is human increases.

[0044]

[0045] Here, A is the maximum signal magnitude in the Doppler spectrum of the current frame, A' is the maximum signal magnitude in the Doppler spectrum of the previous frame, V is the velocity of the target detected in the current frame, and T up represents the time interval between the above frames.

[0046] As shown in Fig. 1, since the received radar signal attenuates significantly with distance, the change in signal magnitude (feature vector 3) must be normalized by the detected distance. However, in the case of a Doppler radar, since the distance value is unknown, the difference in signal magnitude is [calculated as in Equation 1] as a value corresponding to the difference in distance (V×T up The amount of change in signal magnitude can be estimated by compensating through ). The derivation principle of Equation 1 is disclosed in detail in Korean Published Patent No. 10-2023-0080737, which is the background technology of the present invention.

[0047] However, since the short-range detection area within 2 to 3 meters is a section where signal attenuation occurs rapidly depending on the distance, errors may occur when using feature vector 3 for human detection.

[0048] In addition, unlike automobiles, mobile humans have the characteristic that the position of the reflected scattering point continuously changes frame by frame, and the detected speed value may vary slightly depending on the position of the scattering point. Considering this, the performance can be improved in the embodiment of the present invention by additionally considering the amount of speed change (feature vector 4).

[0049] The fourth feature value (L) corresponds to an indicator (feature vector 4) representing the rate of change of velocity over time for the target. This fourth feature value can be obtained through L = |V-V'| / V. Here, V represents the velocity value corresponding to the maximum signal magnitude within the Doppler spectrum of the current frame, and V' represents the velocity value corresponding to the maximum signal magnitude within the Doppler spectrum of the previous frame.

[0050] FIG. 4 is a diagram showing an example of the Doppler spectrum of the immediate past frame and the current frame for a moving pedestrian in an embodiment of the present invention, and FIG. 5 is a diagram showing an example of the Doppler spectrum of the immediate past frame and the current frame for a moving vehicle in an embodiment of the present invention.

[0051] In the case of the pedestrian in Fig. 4, it can be seen that the pedestrian's Doppler scattering points occur in various ways, and the shape of the spectrum is also unpredictable despite being a continuous series of frames. On the other hand, in the case of the vehicle in Fig. 5, it can be seen that the shape of the Doppler spectrum is distributed in a very sharp manner, and the shape (magnitude and velocity components) appears very similar even in a continuous series of frames.

[0052] In addition, in the case of the pedestrian in Fig. 4, the position of the scattering point of the maximum signal size changes even in a very short period of time across consecutive frames, and the corresponding speed value is also slightly different. However, in the case of the vehicle in Fig. 5, the position of the scattering point of the maximum signal size and the corresponding speed value are almost similar across two consecutive frames.

[0053] As such, in the case of the moving human in Fig. 4, there is almost no change in the maximum signal size (signal peak value) between the previous frame and the current frame, but it can be seen that a change in the speed value corresponding to the maximum signal size is observed between the previous frame and the current frame. In the case of the moving vehicle in Fig. 5, there is almost no change in the maximum signal size (signal peak value) between the previous frame and the current frame, but it can be seen that the change in the speed value corresponding to the maximum signal size is also not significant.

[0054] Therefore, it can be seen that the rate of change of the velocity value of the peak signal detected in the current frame compared to the previous frame is also an important characteristic for distinguishing moving objects as humans and vehicles.

[0055] Considering these points, in an embodiment of the present invention, a fourth feature value (L), which is the rate of change of the speed value corresponding to the maximum signal size of the current frame compared to the previous frame, may be additionally considered to more accurately identify a moving human.

[0056] Next, the human determination unit (130) can identify the presence or absence of a moving human using multiple feature values. At this time, the human determination unit (130) can classify the presence or absence of a moving human by applying multiple feature values ​​to a pre-trained machine learning model. The machine learning model can be pre-trained based on big data collected under various human and environmental conditions and can be implemented including decision trees, SVM models, linear regression models, DNN models, etc.

[0057] Here, the human determination unit (130) may determine the presence or absence of a moving human by utilizing all of the multiple feature values ​​(M, N, K, L), or may determine the presence or absence of a moving human by utilizing some of the selected features. In this way, the human determination unit (130) may determine the presence or absence of a moving human by appropriately selecting and combining the four feature values ​​depending on the situation.

[0058] The human determination unit (130) can identify the presence or absence of a moving human by operating a first mode that utilizes all of the first to fourth feature values ​​(M, N, K, L) according to the set situation conditions, or by operating a second mode that utilizes only the first, second, and fourth feature values ​​(M, N, L) excluding the third feature value (K). At this time, the applied mode may be pre-set, selected by the user, or modified depending on the situation.

[0059] In an embodiment of the present invention, the human determination unit (130) may be configured to operate in one of two modes depending on the detection purpose or condition. If the moving human detection technology of the present invention is to be applied to a short-range detection area within a preset threshold distance (e.g., within 2m), the second mode is executed to exclude the third feature value (K), which may be unreliable at short range, and the remaining first, second, and fourth feature values ​​(M, N, L) are applied to a machine learning model to identify whether the object is a moving human (pedestrian) or a vehicle.

[0060] Additionally, when the human determination unit (130) intends to apply the mobile human detection technology of the present invention to a long-distance detection area of ​​a pre-set threshold distance or greater (e.g., 2m or more), it can execute a first mode to apply the first to fourth feature values ​​(M, N, K, L) to a machine learning model to identify whether a mobile human is present. Here, the mode switching according to situational conditions can be performed by a control unit (not shown).

[0061] In this way, the human determination unit (130) can determine whether a moving human exists by applying respective feature values ​​to a decision tree, an SVM model, a linear regression model, a DNN model, etc. In addition, the machine learning model may be constructed individually for two modes.

[0062] FIG. 6 is a diagram illustrating a method for identifying moving humans using a Doppler radar sensor according to an embodiment of the present invention.

[0063] First, the mobile human identification device (100) obtains the Doppler spectrum for each target detected in each frame from the radar reception signal (S610).

[0064] And, the mobile human identification device (100) analyzes the number of Doppler bins greater than a reference value in the Doppler spectrum of each frame for the target, the maximum signal magnitude in the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude to extract multiple feature values ​​related to the movement of the target (S620).

[0065] More specifically, the mobile human identification device (100) can extract first to fourth feature values ​​by calculating the number of Doppler bins (M) greater than a reference value in the Doppler spectrum of the current frame for the target, the difference (N) between the first feature value (M') of the previous frame and the first feature value (M), the rate of change (K) of the maximum signal magnitude of the current frame relative to the previous frame, and the rate of change (L) of the velocity value corresponding to the maximum signal magnitude of the current frame relative to the previous frame.

[0066] Next, the mobile human identification device (100) identifies the presence or absence of a mobile human using a plurality of feature values ​​(S630). At this time, the mobile human identification device (100) may determine the presence or absence of a mobile human by applying the first to fourth feature values ​​(M, N, K, L) to a machine learning model according to the set situation conditions, or it may determine the presence or absence of a mobile human by applying the first, second, and fourth feature values ​​(M, N, L) to a machine learning model.

[0067] According to the present invention, based on a simple and miniaturizable Doppler radar, it is possible to clearly distinguish whether a currently detected moving target is human or not, and based on this, accurately determine the presence of a human.

[0068] The present invention has been described with reference to embodiments illustrated in the drawings, but this is merely illustrative, and those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be determined by the technical spirit of the appended claims. Explanation of the symbols

[0069] 100: Mobile human recognition device 110: Signal acquisition unit 120: Signal processing unit 130: Human determination unit

Claims

Claim 1 A mobile human recognition device based on a Doppler radar, comprising: a signal acquisition unit that acquires a Doppler spectrum for each target detected in every frame from a radar reception signal; and a signal processing unit that extracts multiple feature values ​​related to the movement of a target by analyzing the number of Doppler bins greater than or equal to a reference value, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude in the Doppler spectrum of each frame for the target. A moving human recognition device comprising a human determination unit that identifies the presence or absence of a moving human using the plurality of feature values, wherein the signal processing unit calculates, respectively, a first feature value (M) which is the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of the current frame for a target, a second feature value (N) which is the difference between the first feature value (M') of the previous frame and the first feature value (M), a third feature value (K) which is the rate of change of the maximum signal magnitude of the current frame relative to the maximum signal magnitude of the previous frame, and a fourth feature value (L) which is the rate of change of the speed value corresponding to the maximum signal magnitude of the current frame relative to the speed value corresponding to the maximum signal magnitude of the previous frame, and wherein the human determination unit operates to identify the presence or absence of a moving human by driving a first mode that utilizes all of the first to fourth feature values ​​for a long-range detection area greater than a preset threshold distance, and to identify the presence or absence of a moving human by driving a second mode that utilizes the first, second, and fourth feature values ​​excluding the third feature value for a short-range detection area within the threshold distance. Claim 2 delete Claim 3 A mobile human recognition device according to claim 1, wherein the third feature value (K) is calculated by the following mathematical formula. (Here, A is the maximum signal magnitude in the Doppler spectrum of the current frame, A' is the maximum signal magnitude in the Doppler spectrum of the previous frame, V is the velocity of the target detected in the current frame, T up represents the time interval between the above frames.) Claim 4 delete Claim 5 delete Claim 6 A mobile human recognition device according to claim 1, wherein the human determination unit determines whether a mobile human is present by applying the first to fourth feature values ​​or the first, second, and fourth feature values ​​to a pre-trained machine learning model according to the first mode or the second mode. Claim 7 A method for recognizing a moving human based on a Doppler radar comprises: a step of acquiring a Doppler spectrum for each target detected in each frame from a radar received signal; and a step of extracting multiple feature values ​​related to the movement of a target by analyzing the number of Doppler bins greater than or equal to a reference value, the maximum signal magnitude within the Doppler spectrum, and the velocity value corresponding to the maximum signal magnitude in the Doppler spectrum of each frame for the target. A method for recognizing a moving human, comprising the step of identifying the presence or absence of a moving human using the plurality of feature values, wherein the step of extracting the plurality of feature values ​​calculates, respectively, a first feature value (M) which is the number of Doppler bins greater than or equal to a reference value in the Doppler spectrum of the current frame for a target, a second feature value (N) which is the difference between the first feature value (M') of the previous frame and the first feature value (M), a third feature value (K) which is the rate of change of the maximum signal magnitude of the current frame relative to the maximum signal magnitude of the previous frame, and a fourth feature value (L) which is the rate of change of the speed value corresponding to the maximum signal magnitude of the current frame relative to the speed value corresponding to the maximum signal magnitude of the previous frame, and the step of identifying the presence or absence of a moving human operates by driving a first mode that utilizes all of the first to fourth feature values ​​for a long-range detection area greater than a preset threshold distance to identify the presence or absence of a moving human, and by driving a second mode that utilizes the first, second, and fourth feature values ​​excluding the third feature value for a short-range detection area within the threshold distance to identify the presence or absence of a moving human. Claim 8 delete Claim 9 In claim 7, the third feature value (K) is a moving human recognition method calculated by the following mathematical formula. (Here, A is the maximum signal magnitude in the Doppler spectrum of the current frame, A' is the maximum signal magnitude in the Doppler spectrum of the previous frame, V is the velocity of the target detected in the current frame, T up represents the time interval between the above frames.) Claim 10 In claim 7, the step of identifying the presence or absence of the mobile human is a method for recognizing a mobile human by applying the first to fourth feature values ​​or the first, second, and fourth feature values ​​to a pre-trained machine learning model according to the first mode or the second mode to determine whether the mobile human is present.

Images